Paleoecología del Holoceno en la Gran Sabana, SE Venezuela: Análisis preliminar de polen y microcarbones en la Laguna Encantada
5.3. Materials and Methods
sedges such as Bulbostylis and Rhyncospora; woody elements are scarce and rarely emerge above the herb layer (Huber, 1995b). According to Huber (1994), there is a special type of plant formation (locally called morichal) where the herbaceous stratum remains ecologically dominant (treeless savanna), but the palm Mauritia flexuosa forms characteristic monospecific stands. This association is especially common around lakes, and in the bottom of river valleys and flooded depressions of the southern GS, up to about 1000 m elevation (Huber, 1995b).
Most GS forests are considered to fall within the category of lower montane forests (also called submesothermic forests, between 800 and 1500 m elevation), because of their intermediate position between lowland and highland forests (Hernández, 1999). Common genera include: Virola (Myristicaceae), Protium (Burseraceae), Tabebuia (Bignoniaceae), Ruizterania (Vochysiaceae), Licania (Chrysobalanaceae), Clathrotropis (Fabaceae), Aspidosperma (Apocynaceae), Caraipa (Clusiaceae), Dimorphandra (Caesalpiniaceae), Byrsonima (Malpighiaceae), etc., and their composition varies with elevation (Huber, 1995b).
The GS shrublands usually occur between 800 and 1500 m elevation and are more frequent at the northern area than at the southern part (Huber, 1995b), where our study site is located.
The more common shrub genera are: Euphronia (Euphroniaceae), Bonyunia (Loganiaceae), Bonnetia and Ternstroemia (Theaceae), Clusia (Clusiaceae), Gongylolepis (Asteraceae), Macairea (Melastomataceae), Humiria and Vantanea (Humiriaceae), Ochthocosmus and Cyrillopsis (Ixonanthaceae), Thibaudia, Notopora and Befaria (Ericaceae), Spathelia (Rutaceae), Byrsonima (Malpighiaceae), etc. They usually grow on a rocky, sandstone substrate or deep white sands of alluvial origin (Huber, 1995b).
The GS region is the homeland of the Pemón indigenous group, from the Carib‐speaking family. Today they are sedentary, living in small villages, usually in open savannas. Though the GS population density is currently low, the indigenous settlements have experienced an expansion since the arrival of European missions, and today, more than 17,000 people live in the GS (Medina et al., 2004). Fire is a key component of the Pemón culture as they use it every day to burn wide extensions of savannas, and occasionally, forests (Kingsbury, 2001). The reasons for the extent and frequency of these fires are related to activities such as cooking, hunting, fire prevention, communication, magic, etc (Rodríguez, 2004; 2007). Surprisingly, land use practices, such as extensive agriculture or cattle raising, typical of other cultures strongly linked to fire, are not characteristic of the Pemón culture (Rodríguez, 2004). According to the available evidence, the Pemón seem to have reached the GS very recently, probably around 300 years ago (Thomas, 1982; Colson, 1985), but an early occupation could not be dismissed.
There is some archaeological evidence consisting of pre‐Hispanic remains (spearheads and bifacial worked knives) similar in style to others from about 9000 years old found at other Venezuelan localities (Gassón, 2002; Rostain, 2008). In addition, palynological evidence indicating the occurrence of intense and extensive fires during the Younger Dryas (around 12,400 cal yr BP), suggested a potential early human occupation of the GS (Montoya et al., 2011). However, a definitive assessment is not yet possible. December to March (Huber, 1995a). The lake is within a treeless savanna landscape, surrounded by scattered morichal patches. In the absence of a known local name for the lake, it will be called Lake Chonita for the purposes of the present study, to be consistent with previous studies developed at the same site (Montoya et al., 2011).
The core (PATAM1 B07; 4.67 m long) was obtained in the deepest part of the lake (3.13 m water depth), using a modified Livingstone squared‐rod piston core (Wright et al., 1984). The present study is focused on the detailed analysis, and paleoecological interpretation, of the last three millennia interval (0.03 to 0.97 m). A total of 9 samples were taken along the whole core for radiocarbon dating, 3 of them falling within the interval discussed here. Samples were pretreated using standard acid‐base‐acid procedures (Abbott and Stafford, 1996) and measured at the AMS Radiocarbon Laboratory of the University of California, Irvine (UCI) and Beta Analytic (Beta). Calibration was made using CALIB 6.0.1 and the IntCal09.14c database (http://calib.qub.ac.uk./calib/, last accessed on April 2010).
Twenty‐eight volumetric samples (2 cm3) were taken in the section studied, at 2‐5 cm intervals, for pollen analysis. These samples were processed using standard palynological techniques slightly modified according to the sediment nature (Rull et al., 2010b), after spiking with Lycopodium tablets (batch 177745, average 18,584 ± 1853 spores/tablet). The slides were mounted in silicone oil without sealing. Pollen and spore identification was made according to Hooghiemstra (1984), Roubik and Moreno (1991), Tryon and Lugardon (1991), Herrera and Urrego (1996), Rull (1998a; 2003) and Colinvaux et al. (1999). Counts were conducted until a minimum of 300 pollen and spores were tabulated (excluding Cyperaceae and aquatic plants:
Myriophyllum, Sagittaria and Utricularia), but counting continued until the saturation of diversity was reached (Rull, 1987). Final counts averaged 533 grains per sample. Pollen taxa were grouped according to the vegetation types previously described (Huber, 1986; 1989;
1994; 1995b; Huber and Febres, 2000). All identified pollen taxa were included into the pollen sum, except for Cyperaceae and the aquatic plants mentioned above. Pollen diagrams were plotted with PSIMPOLL 4.26, using a time scale derived from an age‐depth model based on radiocarbon dating, developed with the clam.R statistical package (Blaauw, 2010). The pollen zonation was performed by Optimal Splitting by Information Content (OSIC), and the number of significant zones was determined by the broken‐stick model test (Bennett, 1996). Only pollen types over 0.4% were considered for zonation. Interpretation was based on comparison with modern samples from previous studies (Rull, 1992, 1999) and the known autoecology of taxa found (Marchant et al., 2002; Rull, 2003). NPP were analyzed on pollen slides, and plotted in percentages based on pollen sum. NPP identification was made according to Montoya et al.
(2010) and literature therein. Charcoal counts were carried out in the same pollen slides, considering two size classes (Rull, 1999):
- Type I (smaller microcharcoal particles: 5 ‐ 100 μm): used as proxy for mostly regional fires, because of their easy dispersion by wind.
- Type II: (larger microcharcoal particles: >100 μm): used as proxy for local fires.
Bulk density (BD) was measured on 1 cm3 samples, taken every 5 cm down‐core. The samples were weighed wet, and again after drying in a 60ºC oven for 24 hours. Total organic matter was measured every 5 cm by loss‐on‐ignition (LOI) at 550°C (Dean, 1974). There is no measurable calcium carbonate in the sediments, based on LOI measurements made after burning at 1000°C. Magnetic susceptibility (MS) was measured at a 0.5 cm interval using a Tamiscan high‐resolution surface scanning sensor connected to a Bartington susceptibility meter at the University of Pittsburgh.
5.4. Results
5.4.1. Lithology and chronology
The lacustrine sequence is characterized, in the studied section, by dark‐brown organic‐rich sediments. The upper part of the section is characterized by slightly higher magnetic susceptibility values than the lower one, with a major peak between ~30‐24 cm and a minor peak in the upper 10 cm (between 8.5 and 1.5 cm). The organic matter data shows changes in the relative amounts of organic matter and terrigenous (mineral) sediments in the core (Fig.
5.2). Generally, sections with high dry bulk density (Fig. 5.2) also have lower organic matter and high in terrigenous sediments. Dry BD shows a high variability and presents its maximum values between ~30‐24 cm, coinciding with the major MS peak. Organic matter is characterized by a fluctuating trend followed by an abrupt increase in the upper 20‐10 cm of the record.
Figure 5.2. Core stratigraphy, with radiocarbon ages (in radiocarbon years, uncalibrated ages) and sediment description; pollen zones, physical parameters curves and age‐depth model of the study section. MS: Magnetic susceptibility.
The results of AMS radiocarbon dating were used to produce an age‐depth model for the sequence (Table 5.1). The best fit was obtained with a smooth‐spline function (Blaauw, 2010), and is represented in Figure 5.2 only for the interval of interest of this study. The sedimentation rate of the whole sequence varies between 0.02 and 0.17 cm yr‐1. For the interval studied here, the sedimentation rate ranges between 0.02 and 0.08 cm yr‐1. The time interval between samples ranges from 60 to 150 years.
Table 5.1. AMS radiocarbon dates used for the age‐depth model for the whole record. Asterisks mark the dates included in the interval under study. The estimated ages have been extracted from the calibrated ages (WA:
Weighed average).
Laboratory Sample Depth
(cm)
Sample Material Age (yr C14 BP)
Age (cal yr BP) 2σ
Age (cal yr BP) estimation(WA) Beta ‐ 279600* PATAM1_B07/3 13 Pollen extract 890 ± 40 731 ‐ 915 800
Beta ‐ 277185* PATAM1_B07/11 51 Pollen extract 2850 ± 40 2855 ‐ 3078 2730 Beta ‐ 277184* PATAM1_B07/22 98 Pollen extract 3340 ± 40 3471 ‐ 3643 3660
UCI ‐ 43705 PATAM1_B07/32 144 Wood 4080 ± 40 4497 ‐ 4655 4640
UCI ‐ 43706 PATAM1_B07/49 212 Wood 6465 ± 25 7323 ‐ 7403 7380
Beta ‐ 277186 PATAM1_B07/70 298 Pollen extract 9590± 60 10,738 ‐ 11,164 10,690
UCI ‐ 43537 PATAM1_B07/87 362 Wood 9720 ± 70 11,063 ‐ 11,251 11,380
Beta ‐ 247284 PATAM1_B07/93 392 Wood 10,440 ± 40 12,128 ‐ 12,530 12,340 UCI ‐ 43614 PATAM1_B07/99 402 Wood 11,005 ± 45 12,699 ‐ 13,078 12,740
5.4.2. Palynological zonation
The pollen diagram is dominated by pollen assemblages from two different herbaceous plant formations: a treeless savanna, with a nearby forest in the lower part; and a savanna with morichal, coinciding with a decrease in forest elements, in the upper half (Figure 5.3). The pteridophyte spores are not very abundant, though psilate triletes and monoletes are better represented than others. Regarding NPP, Botryococcus, Coniochaeta cf. ligniaria and Neorhabdocoela oocites are the more abundant (Figure 5.4). The stratigraphic variations of the pollen and spore assemblages allowed subdivision of the diagram into two zones:
Figure 5.3. General pollen diagram expressed in percentages. Solid lines represent x10 exaggeration.
5.4.2.1. LCH‐I (97 ‐ 37 cm, 14 samples)
The pollen assemblage is clearly dominated by Poaceae, which presents fluctuating values ranging from 40 to 70% of the total pollen sum, followed by trees (mainly Urticales) (Fig. 5.3).
Some forest elements are also present at high to medium abundances, as for example Urticales (the more abundant of them), Alchornea, Byrsonima, Cecropia, Euphorbiaceae‐type, Miconia, Myrsine and Weinmannia. Mauritia appears at the top of the zone, though with low abundance. The percentages of pteridophyte spores are low, but a slight increasing trend can be observed in psilate monoletes and psilate triletes at the top of the zone. Smaller charcoal particles (5 ‐ 100 µm) remain at low abundances, with an increase at the top of the zone, coinciding with the first appearance of larger particles (> 100 µm). Regarding influx index, it can be observed that all proxies studied show local peaks of different magnitude at the top of the zone. Among aquatic elements (algal remains and aquatic or semi‐aquatic plants; Figure 5.4), Botryococcus is the dominant, with strong fluctuations in its concentration and a sharp decrease in the upper part of the zone. Type 91 (HdV. 91) shows an increase at the upper part of the zone, and Spirogyra peaks at the top. Cyperaceae are also abundant, with minor variations, and Sagittaria shows a slightly decreasing trend at the upper part of the zone.
Regarding fungal spores and other NPP, the more abundant are Coniochaeta cf. ligniaria, Neorhabdocoela oocites, Cercophora‐type and Sordaria‐type, thought Sordariales also presents a peak at the lower part of the zone (Fig. 5.4).
Figure 5.4. General non‐pollen palynomorphs (extra pollen sum taxa) diagram expressed in percentages respect to pollen sum. Solid lines represent x10 exaggeration. HdV: Hugo de Vries Lab; IBB: Institut Botànic de Barcelona.
5.4.2.2. LCH‐II (37 ‐ 3 cm, 14 samples)
The pollen assemblage is marked by an abrupt increase of Mauritia likely at the expense of trees, in the lower half of the zone, and of trees and Poaceae in the upper part, from around 35 cm upwards (Fig. 5.3). There is a decrease of Mauritia and a return to the former higher values of Poaceae in the intermediate part of the zone (32 ‐ 18 cm). Above this depth, Mauritia increases again synchronously with a decrease in Poaceae. There is a general decreasing trend of nearly all the forest elements, which in some taxa, as Alchornea and Bonyunia‐type, represent almost their complete disappearance. Pteridophyte spores remain at similarly low values to the previous zone. Psilate triletes has higher values at the base of the zone, showing a slightly decreasing trend from ~30 cm upwards. Smaller charcoal particles maintain the increasing trend initiated at the upper part of the previous zone, and experience three abrupt peaks, the first one around 32 to 23 cm, and the other two, of higher magnitude, at 18 and 8 cm, respectively (Fig. 5.3). Larger charcoal particles remain low at the beginning of the section, and show a pattern similar to smaller particles, but significantly lower in magnitude, throughout the zone. All the biological proxies analyzed show an increase in their influx indices, except for pteridophyte spores. Aquatic elements and fungal spores and other NPP are characterized by lower abundances respect to the former zone, except for Cyperaceae, which show similar values (Figure 5.4). The correlation between Mauritia and total charcoal influx index curves (Figure 5.5) was performed, obtaining an R value of 0.718, which is significant for α < 0.001.
Figure 5.5. Influx index diagram of Mauritia pollen and total charcoal particles. Solid lines represent x10 exaggeration.
5.5 Discussion
The region around Lake Chonita has remained a savanna during the last three millennia, but a significant vegetation change occurred around 2000 years ago. Indeed, prior to 2180 cal yr BP, a treeless savanna landscape with nearby forests dominated the site, but the last two millennia have been characterized by forest retraction and the establishment of a morichal, which remains until present. The paleoecological sequence is discussed in the following sections in the context of northern South American savannas, and the contribution of these results to the understanding of the fire‐vegetation relationships at South GS.
The sedimentary features and the presence of aquatic organisms indicate that the lake probably was already established prior to 3640 cal yr BP. The pollen assemblage of this zone indicates a treeless savanna landscape without morichales. The abundance of forest elements suggests that this formation was probably closer and/or more expanded than today. The continued presence of smaller charcoal particles ‐indicative of regional fires‐ together with the continuous presence of Cecropia ‐a secondary colonizer‐ may indicate some regional fires of low intensity occurred. The lack of coarse charcoal indicates local fires did not occur. The first appearance of larger microcharcoal particles, as proxies for local fires, were recorded at ~2400 cal yr BP. This occurred synchronously with the first appearance, though at low values, of Mauritia pollen, and an increase in psilate triletes. These spores have been related with early stages of secondary succession after fire, in other sites of the GS (Rull, 1999). The high values of Botryococcus and Neorhabdocoela oocites from 3640 to 2800 cal yr BP suggest that lake levels were stable. During this time period climate might have varied from a higher water balance prior to 2800 cal yr BP to lower moisture availability from this date to the end of the interval, as indicated by the lower values of aquatic organisms, mainly Botryococcus and Neorhabdocoela. This is in agreement with the Encantada record (Montoya et al., 2009), but it does not coincide exactly with other GS records, as for example DV or ST (Rull, 1992). Dating inconsistencies in previous records derived from the use of large quantities of bulk sediment for dating using conventional radiocarbon methods instead of AMS techniques, and the few dates available for a sound age‐depth model can not be dismissed for this time interval.
Similar trends regarding water levels and climate have been observed in some paleoecological and paleoclimatic records from northern South America. For instance, Lake Valencia (Fig. 5.1) had higher water levels from 6000 to 3000 14C yr BP (~6840 to 3200 cal yr BP), except for a short interval of lower lake levels centered at 3300 14C yr BP (~3550 cal yr BP) (Bradbury et al., 1981; Leyden, 1985; Curtis et al., 1999). From this, some of these authors inferred a high precipitation/evaporation ratio (P/E) determined by higher insolation and changes in the latitudinal position of the Intertropical Convergence Zone (ITCZ) (Curtis et al., 1999). Haug et al. (2001) inferred a decrease in precipitation from 5350 cal yr BP in the Cariaco Basin (Fig.
5.1), with large century‐scale variations between ~3750 to 2750 cal yr BP. In the Colombian Llanos Orientales, a wetter interval was suggested for the middle Holocene, peaking around 4000 cal yr BP (Marchant and Hooghiemstra, 2004). Such climatic inferences were supported by evidence of forest expansion in different records (e.g. Behling and Hooghiemstra, 1998, 1999, 2000; Berrío et al., 2002). Contrarily, the Rupununi savannas of Guyana (Fig. 5.1), would have had a continuous presence of treeless savanna since the middle Holocene, with an increase in Poaceae around 3000 14C yr BP (~3200 cal yr BP) (Wymstra and van der Hammen, 1966).
Therefore, a likely forest expansion in the present savanna areas of northern South America the general GS landscape continued to be dominated by treeless savannas. The sudden increase of Mauritia coincides with a decrease of Poaceae and forest elements. While Poaceae abundance returned to former values at ca. 1920 cal yr BP, the forest did not show any recovery until the present. The increase in fire incidence during this interval could have been decisive in this sense, favoring the establishment of morichal communities, as suggested by several former studies (Rull, 1992, 1998b, 1999; Montoya et al., 2009). The potential establishment of a drier regional climate since 2800 cal yr BP (Bradbury et al., 1985; Curtis et al., 1999; Berrío et al., 2000, 2002; Behling and Hooghiemstra 2001; Wille et al., 2003), might indicate some level of climatic influence (or a synergistic fire‐climate coupling) on forest retraction. The treeless savanna expanded again from 1920 to 1120 cal yr BP, synchronously with a decrease of Mauritia abundance. At the same time, there is a major peak in MS and BD curves. Such synchrony could be interpreted as a higher input of terrigenous sediments to the watershed due to erosion processes caused by the existence of a more open landscape resulting from Mauritia clearing. After that, two major charcoal peaks recorded at ca. 1120 and 480 cal yr BP coincide with the morichal expansion. Thus, it is suggested that the present‐
day landscape around Lake Chonita was established around 1120 cal yr BP. The MS minor peak occurred this time paralleled the Mauritia increase and is dated ca. 500 to 50 cal yr BP, which is synchronous with the Little Ice Age (LIA) recorded in the Venezuelan Andes, as a cool and humid interval linked to solar activity cycles (Polissar et al., 2006). In Lake Chonita, the only potential evidence for more humid conditions is the Mauritia increase at the top. However, aquatic elements indicate that during the whole interval moisture conditions were more or less stable, and similar to present‐day, with minor variations, so a definitive interpretation can not be made.
The recent appearance and sudden increase of Mauritia, or the establishment of present‐day morichales, coinciding with an increased fire incidence have also been reported in most sequences in the GS (e.g.: DV, ST, Urué and Encantada) (Rull, 1992, 1999; Montoya et al., 2009). Sudden increases of Mauritia and/or slightly drier climate than mid Holocene relative to the last millennia have also been reported in several studies in nearby areas. In the Venezuelan Llanos, Mauritia presence was also reported only for the last two millennia, in a climate likely more humid than the previous interval (Leal et al., 2002, 2003). In the Colombian Llanos, the same trends have been observed, during the last two millennia, in several localities (Behling and Hooghiemstra, 1998, 1999, 2000, 2001; Berrío et al., 2000, 2002; Wille et al., 2003).
Hence, there is a general agreement regarding the influence of increased human impact, usually through fire, in the establishment of morichales, and the shaping of the present savanna landscapes during the last two millennia.
5.5.2. Mauritia, climate, fire, and human occupation in the GS
Several studies developed in the GS have revealed the continuous persistence of savannas since at least the early Holocene (Rull, 2007; Montoya et al., 2011). However, the taxonomic composition of this biome has shown the dynamic nature of its plant communities.
This is the case of morichales, whose occurrence has been traditionally considered indicative of warm and wet lowlands (and midlands) of northern South America. As a consequence, morichal expansions observed in paleoecological records have been generally interpreted in terms of wetter climate (e.g. Rull, 1992, 1999; Behling and Hooghiemstra, 1999; Berrío et al., 2000; Leal et al., 2002, 2003). The appearance and expansion of Mauritia‐dominated communities in the Colombian Llanos Orientales (Fig. 5.6), likely agree with this climatic interpretation. Thereby, this palm was recorded for first time around the middle Holocene, where a wet period was documented for the region (Marchant and Hooghiemstra, 2004). In our study, however, the recent morichal expansion occurred in a climate drier than the preceding phase which, at first, seems contradictory. Nevertheless, this evidence also suggests that climate is not the only factor affecting the morichal occurrence and distribution at GS, which appears to be linked more strongly to fire incidence or to fire‐climate synergies (Fig.
5.5). The synchrony between increased fire frequency and morichal establishment recorded in several GS sequences together with the correlation degree obtained, as well as the common presence of charcoal particles in palynological slides, supports this view. This is in agreement
5.5). The synchrony between increased fire frequency and morichal establishment recorded in several GS sequences together with the correlation degree obtained, as well as the common presence of charcoal particles in palynological slides, supports this view. This is in agreement